RE: Are duplicate property/value pairs permitted for a resource?

Or, if such a processor permitted an agent to see only its own assertions,
then essentially every agent has its on separate RDF dataset, and every
dataset is self-consistent.  Such a processor would be pretty limiting,
however, since data could not be shared between agents.

-----Original Message-----
From: Samuel Yang [mailto:syang@peoplemoverinc.com]
Sent: Friday, April 09, 1999 3:48 PM
To: 'John Cowan'; Samuel Yang
Cc: 'www-rdf-comments@w3.org'
Subject: RE: Are duplicate property/value pairs permitted for a
resource?


John Cowan wrote:

> The problem is what happens when
> two different agents have contradictory *facts*.   One of them
> is wrong, obviously, but there is no definition about what RDF
> does in such circumstances.

> This is why I say that the semantics of deletion matters.
> What does it mean to remove a statement from the database
> of RDF facts?  There is no notion in RDF of *authoritative*
> claims, and probably there should be.

I think "facts" must be incontrovertible within a single RDF dataset.
Ideally a single RDF dataset should be shared only by agents that share the
same belief system.  However, if a single dataset is shared between agents
with different belief systems, then those agents somehow deemed "unreliable"
(perhaps all of them) must use reified statements.  Otherwise, the dataset
is itself unreliable and therefore useless.

In fact, if as John proposed, an RDF processor were able to support
alternative beliefs as facts, the processor would still have to internally
represent every fact as a reified statement.  That is, the processor must
remember internally for each fact which agent asserted it.  But the
knowledge of who asserted what needs to be made available to users of the
processor so that the RDF data can be used by them in any meaningful way.
In other words, we are basically back to just using reified statements.

Received on Friday, 9 April 1999 19:00:18 UTC